Shopify’s CTO Mikhail Parakhin highlighted the company’s significant advances in AI adoption, stating that Shopify has achieved nearly universal internal AI tool usage, thanks to a pivotal inflection in model quality in December. This transition has led to the emergence of innovative tools such as Tangle, Tangent, and SimGym, which facilitate reproducible experimentation, customer simulation, and optimization loops. Parakhin emphasized that as coding productivity with AI improves, the main bottlenecks have shifted from code generation to the areas of pull request reviews and CI/CD stability, illustrating a fundamental evolution in the software development process at Shopify.

SimGym: SimGym is Shopify’s AI-powered simulation environment that models buyer behavior using historical merchant data and browser-based agents for realistic offline A/B testing. Launched recently to eligible merchants, it analyzes live or draft themes, predicts conversion changes, and suggests optimizations like layout adjustments. Its strength lies in leveraging Shopify’s vast data moat for accurate, category-specific customer trajectories.
Tangle: Tangle is Shopify’s open-source workflow engine designed for reproducible and collaborative ML experimentation and data processing at scale. It uses content-addressed caching for efficiency across teams, differing from schedulers like Airflow by enabling seamless development-to-production transitions. In recent discussions, it underpins auto-research and experimentation, fostering network effects in shared pipelines.
Shopify: Shopify is a commerce platform powering millions of online stores worldwide with tools for building, managing, and scaling e-commerce operations. The company has recently achieved near-universal AI tool adoption among its workforce, shifting engineering practices toward AI-native development with unlimited high-end model access. This is enabling innovations like advanced customer simulation and automated experimentation to enhance merchant experiences.
Tangent: Tangent is an agentic auto-research tool built atop Tangle that iteratively optimizes experiments, prompts, and systems through critique loops and modifications. It democratizes advanced experimentation for product managers and domain experts by automating improvements in areas like search latency and storage. Recent internal use has shown it compounding with other systems for broader engineering gains.
Mikhail Parakhin: Mikhail Parakhin is the Chief Technology Officer at Shopify, directing AI, machine learning, and engineering efforts. Previously, he led key Microsoft units including Bing, Edge, Windows, and advertising; in a recent interview, he discussed Shopify’s AI inflection point, internal tools like Tangle and Tangent, and the shift in development bottlenecks to review and deployment.

AI Adoption: Shopify has reached near-universal internal AI tool usage following a December model-quality inflection, with CLI-based tools outpacing IDE plugins.
Tool Synergy: Tangle, Tangent, and SimGym combine for powerful reproducible experimentation, customer simulation, and optimization loops.
Development Bottlenecks: AI coding productivity has shifted bottlenecks from generation to PR reviews, CI/CD stability, and deployment rollbacks.